Chapter2 LPWANs Background
LPWANs
Background
Chapter2 LPWANs Background
2.1 Introduction
This chapter provides a comprehensive overview of Low Power Wide Area Networks (LPWANs),
a class of wireless communication technologies designed for long-range, low-power data
transmission in IoT systems. It outlines the fundamental characteristics of LPWAN architectures,
including their access-side requirements such as energy efficiency, long-range connectivity, and
uplink-dominated traffic. The chapter also introduces a range of LPWAN technologies—both
standardized and proprietary—and explains their structural components, cloud integration trends,
and approaches to data handling in large-scale IoT deployments.
2.2 What is LPWANs
LPWANs define a wireless network family, which can be considered as part of the third
generation of IoT networking, where the focus is the provisioning of cloud services to connected
things . Fig 2.1 compares the key wireless technology families in terms of data rate, range, and
power consumption. Clearly, the LPWAN technologies are the winners in terms of range and
power consumption. The following are well-established LPWAN technologies: Extended
Coverage-Global System for Mobile communications-IoT (EC-GSM-IoT), Long Term Evolution
Machine type communication (LTE-M), Narrowband-IoT (NB-IoT), MY IoT (MIOTY), Random
Phase Multiple Access (RPMA), Weightless, Wize, Sigfox, and LoRa.
Fig. 2.1 Wireless technology families with respect to data rate, range, and power efficiency.
The first three technologies (i.e., EC-GSM-IoT, LTE-M, and NB-IoT) have been defined by the
3rd Generation Partnership Project (3GPP) and are referred to as Cellular IoT (C IoT)
technologies, whereas the rest have not been standardized by 3GPP. This criterion also splits the
technologies into those that use the licensed part of the spectrum (3GPP technologies) and those
Chapter2 LPWANs Background
that operate in the license-free spectrum (non-3GPP technologies). From an architectural
perspective, LPWANs are composed of five types of nodes: the EDs and the GWs, which com
pose the access-side, and the Network Gateway, the LPWAN Authentication, Authorization, and
Accounting (AAA) Server, and the Application Server, which compose the server-side. EDs are
deployed in massive numbers and tend to broad cast their messages in the hope of being received
by a GW that covers the area. GWs forward these messages to the server-side, where numerous
functions at the network and application layers take place. At the server-side, the Network
Gateway receives this transmitted information and either per forms internal processing or re-
distributes the information to the rest of the interested parties, i.e., the LPWAN AAA Server and
the Application Server. Existence of the LPWAN AAA Server and the Application Server is
optional, because the Network Gateway may be able to handle all the necessary server-side
functionality. Nevertheless, common practices dictate the use of dedicated servers for the
different functionalities and types of messages, e.g., an LPWAN AAA Server may authorize
specific EDs to participate in the network, whereas an Application Server may be the bridging
point between internal server-side application data and external application end user. A growing
trend in server-side implementations is the utilization of a cloud infrastructure. Although the
nomenclature and the full node deployment might be different from one technology to another, an
LPWAN system architecture can be usually described by a structure similar to the one depicted in
Fig. 2.2
Fig. 2.2 A generic LPWAN system architecture.
Chapter2 LPWANs Background
2.3 ACCESS-SIDE CONSIDERATIONS
Focusing on their access-side, LPWANs need to meet certain demands, such as energy efficiency,
long range communication, low-cost operation, and support of Uplink (UL) dominated
connections.
2.3.1 Energy Efficiency
In all LPWAN technologies, the transmission process is recognized as the primary contributor to
energy consumption. Therefore, selecting an efficient radio interface and implementing strategic
transmission planning are critical to prolonging system lifespan. Additionally, the use of
lightweight communication protocols is crucial—not only to reduce the frequency of
transmissions but also to shorten their duration. In certain applications, end devices (EDs) utilize
solar panels for power, adopting an energy harvesting model that shifts a significant portion of
network management responsibilities away from the core system.
2.3.2 Long Range
As implied by the name of the networking paradigm, long-range communication is a fundamental
requirement of LPWAN technologies. Typically, these networks are designed to operate over
distances of several kilometers, generally ranging from one to ten kilometers. In some
experimental configurations, communication over distances up to one hundred kilometers has
been demonstrated. These extended ranges are made possible through the use of resilient Physical
Layer (PHY) protocols. Specifically, ranges around one kilometer represent the standard
operational coverage in urban and suburban environments. In contrast, distances approaching ten
kilometers are achievable in areas with minimal obstructions, such as rural settings with clear
Line of Sight (LOS). Furthermore, although less common, distances nearing one hundred
kilometers have been successfully tested under experimental conditions. A notable example is a
LoRaWAN transmission that reached a record distance of 766 kilometers using a weather
balloon.
Chapter2 LPWANs Background
2.3.3 LowCost
On the access side, gateways (GWs) are typically deployed in small numbers due to the extensive
coverage range of LPWANs. These gateways function primarily as packet forwarders, with most
processing responsibilities delegated to the server side—an approach that helps reduce the overall
cost of gateway infrastructure. In contrast, end devices (EDs) are deployed in large quantities,
which necessitates keeping their development costs low. This is achieved by simplifying their
functionality and minimizing hardware and software requirements. EDs can often be assembled
using basic components such as suitable radio modules combined with widely available
development platforms like Arduino, Raspberry Pi, or BeagleBoard. Additionally, cost efficiency
can be further enhanced by applying optimization strategies on the server side.
2.3.4 UL Dominance
The access layer of LPWANs supports Wireless Sensor Networks (WSNs) and is primarily
characterized by uplink-dominant communication. This means that the majority of data
transmissions occur from end devices (EDs) to gateways (GWs). In practical deployments, EDs
are placed in the field to collect environmental or situational data through their sensors,
transmitting this information to the server side either periodically or in response to specific
events. Unlike traditional WSNs, which often rely on mesh or ad-hoc topologies, LPWANs are
designed around a star topology, where each ED communicates directly with a gateway.
2.4 Cloud Integration in LPWAN Server-Side Operations
On the server side of LPWAN architectures, key components such as the Network Gateway,
Application Server, and AAA Server are essential. The development and deployment of these
components are increasingly supported through open-source software and cloud-based services.
Cloud computing offers flexible and scalable infrastructure, enabling network operators and
service providers to adopt either fixed or pay-as-you-go billing models.
A prominent trend is the rise of Everything-as-a-Service (XaaS), which encompasses various
cloud service models:
Chapter2 LPWANs Background
• Software-as-a-Service (SaaS): Allows users to access provider-developed applications with
limited configurability.
• Platform-as-a-Service (PaaS): Enables deployment of user-developed software within a
managed environment, without control over the underlying hardware.
• Infrastructure-as-a-Service (IaaS): Provides full control over virtual platforms, such as OS and
storage, while abstracting the physical infrastructure.
Emerging service models such as Database-as-a-Service (DBaaS) and Forensics-as-a-Service
(FaaS) highlight cloud computing’s growing role in IoT data management and security.
In the LPWAN context, RPMA is a notable example of a PaaS solution, offering tools for
application development. Other XaaS models provide backend access, infrastructure, and storage
tailored to IoT systems.
Major cloud providers like AWS (IoT Core), Microsoft Azure, Google Cloud, Alibaba Cloud,
IBM Watson, Tencent IoT Hub, and Oracle Cloud offer robust IoT platforms. Alongside them, a
growing number of IoT-focused cloud service providers continue to expand the ecosystem with
specialized support for IoT functionalities
2.5 Data Management in LPWANs
LPWANs belong to the third generation of IoT systems, which emphasize cloud-based data
storage and processing through data-centric and data-driven networking approaches. Within this
context, LPWANs must address the core challenges associated with big data, commonly known
as the five Vs: managing large Volumes of data, handling rapid Velocity, supporting a wide
Variety of formats, extracting Value, and ensuring Veracity or trustworthiness of the data. To meet
these challenges, advanced LPWAN architectures have been developed, often drawing on
established reference models from Industry 4.0, such as the Reference Architectural Model
Industrie 4.0 (RAMI 4.0), the Industrial Internet Reference Architecture (IIRA), the International
Data Spaces Reference Architecture Model (IDS-RAM), and the IoT World Forum (IoTWF)
Reference Model.
Chapter2 LPWANs Background
While the following sections explore LPWANs primarily from a networking perspective—
particularly the communication between end devices (EDs) and gateways (GWs)—this section
shifts focus to the data generated by EDs. These devices collect information via sensors and
forward it to aggregation points. Here, we examine the fundamental stages of data handling in
LPWAN environments: acquisition, storage, analysis, and visualization. More specific
networking concerns, including individual LPWAN technologies and LoRa/LoRaWAN networks,
are addressed in subsequent sections.
2.5.1 Fundamental Data Processes for LPWANs
Data management in LPWAN systems involves four key processes: acquisition, storage,
visualization, and analysis.
Data acquisition is central to sensing and actuation, as end devices (EDs) are primarily
responsible for collecting or responding to data. Two models are commonly used:
• The pull model, where data is requested by the user (typically server-side), exemplified by the
HTTP GET method.
• The push model, where data is automatically sent as it becomes available, using protocols like
MQTT and Webhooks. Notably, OASIS is working on standardizing MQTT for Sensor Networks
(MQTT-SN) due to the rising popularity of Wireless Sensor Networks (WSNs).
Data storage is crucial for system reliability and future analysis. Basic storage can be done using
files in formats such as .json, .csv, or .txt, with JSON and CSV being the most widely used for
their readability. XML and its efficient variant EXI offer semantic compatibility and are useful
for constrained devices. More advanced systems utilize databases, including:
• Document-oriented databases (e.g., MongoDB, DynamoDB),
• Time-series databases (e.g., InfluxDB, Prometheus), and
• Graph databases (e.g., Neo4j, Dgraph), which structure data as nodes and edges representing
communication flows.
Data visualization and analysis turn stored data into actionable insights. Tools like Grafana and
Thing Speak offer open platforms for real-time dashboards and basic processing. Thing Speak
Chapter2 LPWANs Background
uses MATLAB for deeper analysis, while platforms like TagoIO support programming in
JavaScript and Python.
With the evolution of IoT, LPWANs now support integration with Artificial Intelligence (AI) and
Machine Learning (ML). Frameworks such as PyTorch and TensorFlow enable ML deployment
not only in high-performance nodes like gateways or servers but also on constrained devices
through TinyML, making smart local decision-making possible.
2.5.2 Advanced Data-Driven Services on Top of LPWANs
Beyond core data processes, LPWANs—especially in industrial contexts—are increasingly
adopting advanced data-driven models. Two key innovations are:
1. Common frameworks for asset representation, such as the Asset Administration Shell
(AAS), which standardizes data formats and communication protocols (e.g., OPC-UA by the
OPC Foundation).
2. Digital Twinning, where virtual models replicate physical LPWAN components and
processes.
In industrial environments, an asset may refer to any connected device, machine, or process.
AAS enables discovery, access, and description of these assets using standardized APIs and
data models. Discovery can be supported via directories, search engines, or networking
protocols. Data descriptions often use OPC-UA models, which align well with LPWANs due to
their publish/subscribe communication style.
Beyond basic models, Companion Specifications target specific industries, while tools like the
Digital Twin Definition Language (DTDL) support inter-AAS communication.
The Digital Twin (DT) represents a digital version of a physical asset, using real-time data to
monitor and influence its behavior. DTs are foundational to Industry 4.0 and the Industrial IoT.
Solutions like ELIoT simulate LPWAN systems, treating end devices as clients and gateways
as servers. Platforms such as Azure Digital Twins and tools like Node-RED further enable
digital twin implementation and data flow management in LPWAN environments.
Chapter2 LPWANs Background
2.6 LPWAN TECHNOLOGIES
Having laid the foundation of LPWANs, we proceed with a discussion on wireless technologies
that can support such type of networks.
Fig 2.3 Overall Classification of LPWANs by spectrum type and use case similarity.
2.6.1 EC-GSM-IoT
EC-GSM-IoT, introduced in 3GPP Release 13, extends existing 2G/GSM infrastructure to
support long-range IoT communication. Built on enhanced GPRS (eGPRS)/EDGE (2.75G), it
offers better data rates (up to 240 kbps) while maintaining GSM’s wide coverage.
Key enhancements include:
• Extended power efficiency via Power Saving Mode (PSM) and enhanced Discontinuous
Reception (eDRX) of up to 52 minutes (vs. 11 minutes in GSM), significantly increasing device
lifespan.
• Optimized signaling and reduced retransmissions lower complexity and improve battery life.
• Security upgrades enable LTE-level protection, with improvements in authentication, integrity,
and encryption.
EC-GSM-IoT uses cost-effective devices capable of operating for up to 10 years on minimal
power (5 Wh average), transmitting across GSM bands (395–1960 MHz) in 200 kHz channels. It
supports two power classes:
• 23 dBm (MCL: 154 dB)
• 33 dBm (MCL: 164 dB)
Data rates depend on modulation: GMSK supports 0.35–70 kbps, and 8PSK reaches up to 240
kbps. Multiple access methods include CDMA, TDMA, and FDMA, used for both uplink and
downlink communication.
2.6.2 LTE-M
LTE-M, encompassing LTE-MTC, LTE Cat M1, and LTE-M, is a CIoT technology designed for
low-power, wide-area connectivity. It operates within the LTE spectrum and coexists with
existing cellular technologies (2G–5G), benefiting from reduced deployment costs due to
software upgrades on existing LTE base stations.
Security and privacy features are inherited from LTE, including robust mechanisms for data
confidentiality, device authentication, and message integrity. Mobility support and handover
capabilities were enhanced in 3GPP Releases 14 and 15, along with the introduction of Cat. 2
devices, VoLTE, improved latency, and reduced power consumption. Release 16 continued this
evolution by enhancing mobility, coexistence with 5G (NR), and introducing new power-saving
techniques.
LTE-M simplifies LTE functionality to reduce end device (ED) complexity and energy use. It
limits features such as wideband control channels, transmission modes, and Hybrid ARQ,
reflecting its focus on small data transmissions.
Devices operate in low-power modes (e.g., PSM, eDRX, C-DRX) and can last up to 10 years
with an average consumption of 5Wh. LTE-M uses cellular bands (462.5 MHz to 2655 MHz),
depending on local regulations, with a bandwidth of 1.08 MHz. Two power classes are defined
(23 dBm and 20 dBm), both offering an MCL of 155.7 dB.
Data rates are among the highest in LPWANs, reaching 1 Mbps generally, and up to 7 Mbps UL /
4 Mbps DL in Cat. 2 devices using 5 MHz bandwidth. QPSK and 16-QAM are used for
modulation; SC-FDMA handles uplink access, while OFDMA is used in the downlink—
consistent with LTE standards. Performance is further boosted using 15 kHz tone spacing and
Turbo codes.
2.6.3 NB-IoT
NB-IoT is one of the most prominent LPWAN CIoT technologies. It defines a dedicated radio
interface for IoT communications, advancing beyond EC-GSM-IoT and LTE-M, which are based
on GSM and LTE radio interfaces, respectively.
NB-IoT was first specified in 3GPP Release 13, with enhancements introduced in Releases 14,
15, and 16. In Release 14, a new device power class of 14 dBm was introduced, along with
improvements in mobility, power consumption, and latency. The data rate was also increased,
reaching up to 250 kbps for both uplink (in multi-tone deployment) and downlink. Release 15
brought further upgrades, including range extension up to 120 km, support for micro-, pico-, and
femto-cells, and improved wake-up procedures for EDs operating in DRX or eDRX modes.
Release 16 continued to expand the technology, adding new power-saving techniques, improved
support for mobility, directions toward Self-Organizing Networks (SONs), and better coexistence
with 5G NR.
As a 3GPP CIoT technology, NB-IoT operates in the licensed spectrum, both in sub-GHz bands
(e.g., B8, B26) and above 1 GHz (e.g., B66, B74). The system bandwidth is 180 kHz. NB-IoT
supports three modes of operation. In standalone mode, it utilizes GSM spectrum by operating on
a GSM carrier. In guard-band mode, it uses unused resource blocks at the edges of an LTE
carrier, typically reserved for interference mitigation. In in-band mode, NB-IoT occupies an LTE
Physical Resource Block (PRB), corresponding to a 180 kHz bandwidth, since LTE uses 12
subcarriers of 15 kHz each. When connecting, the ED is unaware of the exact operation mode.
In terms of power efficiency, NB-IoT employs the same techniques as LTE-M, namely PSM,
eDRX, and C-DRX. Devices are designed to operate for up to ten years with typical battery
usage of 5 Wh, and further optimizations allow even longer operation. A key advantage of NB-
IoT is its Maximum Coupling Loss (MCL), which can reach up to 164 dB in standalone
deployments, enabling very wide coverage areas.
NB-IoT supports QPSK modulation for both uplink (multi-tone) and downlink, and π/4-QPSK or
π/2-BPSK in single-tone uplink configurations. It uses SC-FDMA in the uplink and OFDMA in
the downlink for multiple access, similar to LTE. To enhance performance, it employs 15 kHz
tone spacing for both uplink and downlink, or 3.75 kHz spacing in single-tone uplink, along with
Turbo coding in the uplink.
2.6.4 MIOTY
MIOTY is a Low Power Wide Area Network (LPWAN) standard developed by the Fraunhofer
Institute under the ETSI Telegram Splitting Ultra Narrow Band (TS-UNB) protocol family,
designed for unlicensed spectrum operation with support for high mobility (up to 120 km/h),
enhanced robustness against co-channel interference, and energy-efficient communication via
Telegram Splitting (TS), which divides messages into sub-packets transmitted pseudorandomly
in time and frequency domains; the gateway reconstructs full messages from partial receptions,
reducing transmission time and improving battery life, with additional support for energy
harvesting; it uses Telegram Splitting Multiple Access (TSMA) for shared channel access,
supports two power classes (14 dBm for 868 MHz in Europe and 23.3 dBm for 915 MHz in
North America), enables end device lifespans up to 20 years, and features a link budget of 154
dB with −139 dBm sensitivity, while offering a data rate around 0.4 kb/s with a 200 kHz
bandwidth; furthermore, MIOTY classifies end devices into Class Z (uplink only), Class A
(uplink with limited downlink), and Class B (supporting multicast and broadcast downlink
commands).
2.6.5 RPMA
RPMA, invented and patented by Myers with On-Ramp Wireless Inc. (now Ingenu), is a Direct
Sequence Spread Spectrum (DSSS) technique that pseudorandomly delays end device (ED)
transmissions to reduce collisions and enhance the Signal to Interference and Noise Ratio
(SINR); unlike typical LPWAN technologies operating in sub-GHz bands, RPMA utilizes the
globally available 2.4 GHz band, offering strong coverage despite lacking the penetration
advantages of lower frequencies, and mitigates interference from technologies like WiFi by
leveraging narrow frequency channels placed between WiFi bands along with high robustness
and its own multiple access scheme; RPMA ensures secure communication via mutual
authentication, end-to-end encryption, and SSL/TLS for backhaul connections; it uses a broader
bandwidth of 1 MHz, with EDs transmitting at up to 21 dBm and base stations (BSs) at 21 dBm
in Europe and 30 dBm in the U.S.; under optimized conditions, EDs can operate for over ten
years—and up to twenty with infrequent, small transmissions; RPMA also boasts a high link
budget of 168 dB, with BS receiver sensitivity at −142 dBm and ED sensitivity at −133 dBm,
and supports data rates of up to 78 kb/s uplink and 19.5 kb/s downlink; since 2018, Ingenu has
offered RPMA as a Platform-as-a-Service (PaaS) solution.
2.6.6 Weightless
The Weightless Special Interest Group (SIG) developed three LPWAN technologies—
Weightless-W, -N, and -P—where Weightless-W operates in the TV White Space (TVWS)
spectrum but faces deployment challenges due to varying global regulations; Weightless-N,
functioning in the sub-GHz unlicensed band and resembling Sigfox, is used by NWave but has
uncertain link budget performance, prompting the SIG to prioritize Weightless-P, which supports
fully acknowledged bidirectional communication with Quality of Service (QoS) mechanisms
such as Forward Error Correction (FEC), Cyclic Redundancy Code (CRC), and Automatic
Repeat Request (ARQ), along with strong security features including AES-128/256 mutual
authentication, message integrity, confidentiality, and secure Over-The-Air (OTA) key exchange;
in terms of deployment, base stations typically transmit at 27 dBm (up to 30 dBm), while end
devices transmit at 14 dBm (also capable of 30 dBm), with idle power consumption at 100 μW
allowing device lifespans of three to eight years; although Weightless-P can theoretically operate
in both licensed and unlicensed sub-GHz bands, it currently functions in unlicensed frequencies
such as 138 MHz, 433 MHz, 470 MHz, 780 MHz, 868 MHz, 915 MHz, and 923 MHz, with
uplink bandwidth options of 12.5 kHz or 100 kHz and downlink channels supporting broadcast,
unicast, and multicast at 100 kHz, 50 kHz, or optionally 12.5 kHz; its maximum link budget is
153 dB under AWGN conditions with −135.5 dBm receiver sensitivity, −17.5 dB SNR, BER of
10⁻⁴, Coding Rate of 12, Spreading Factor of 8, and Offset QPSK (OQPSK) modulation—though
it also supports GMSK modulation with a BT product of 0.3 in both uplink and downlink; for
multiple access, uplink employs a combination of TDMA and FDMA, while downlink uses only
FDMA; maximum data rate reaches 100 kb/s with 100 kHz bandwidth and GMSK modulation,
though uplink performance drops to 10 kb/s with 12.5 kHz bandwidth, and OQPSK delivers
lower data rates in exchange for increased range.
